Spatio-Temporal Analysis of Female Breast Cancer Incidence In
Total Page:16
File Type:pdf, Size:1020Kb
Zhou et al. Chinese Journal of Cancer (2015) 34:13 DOI 10.1186/s40880-015-0013-y ORIGINAL ARTICLE Open Access Spatio-temporal analysis of female breast cancer incidence in Shenzhen, 2007–2012 Hai-Bin Zhou1†, Sheng-Yuan Liu2†, Lin Lei1, Zhong-Wei Chen2, Ji Peng1*, Ying-Zhou Yang1 and Xiao-Li Liu1 Abstract Introduction: Breast cancer is a leading tumor with a high mortality in women. This study examined the spatio-temporal distribution of the incidence of female breast cancer in Shenzhen between 2007 and 2012. Methods: The data on breast cancer incidence were obtained from the Shenzhen Cancer Registry System. To describe the temporal trend, the average annual percentage change (AAPC) was analyzed using a joinpoint regression model. Spatial autocorrelation and a retrospective spatio-temporal scan approach were used to detect the spatio-temporal cluster distribution of breast cancer cases. Results: Breast cancer ranked first among different types of cancer in women in Shenzhen between 2007 and 2012 with a crude incidence of 20.0/100,000 population. The age-standardized rate according to the world standard population was 21.1/100,000 in 2012, with an AAPC of 11.3%. The spatial autocorrelation analysis showed a spatial correlation characterized by the presence of a hotspot in south-central Shenzhen, which included the eastern part of Luohu District (Donghu and Liantang Streets) and Yantian District (Shatoujiao, Haishan, and Yantian Streets). Five spatio-temporal cluster areas were detected between 2010 and 2012, one of which was a Class 1 cluster located in southwestern Shenzhen in 2010, which included Yuehai, Nantou, Shahe, Shekou, and Nanshan Streets in Nanshan District with an incidence of 54.1/100,000 and a relative risk of 2.41; the other four were Class 2 clusters located in Yantian, Luohu, Futian, and Longhua Districts with a relative risk ranging from 1.70 to 3.25. Conclusions: This study revealed the spatio-temporal cluster pattern for the incidence of female breast cancer in Shenzhen, which will be useful for a better allocation of health resources in Shenzhen. Keywords: Breast cancer, Spatial analysis, Spatial autocorrelation, Spatio-temporal clustering Background among women; the ASR of incidence was estimated to be Breast cancer is a leading tumor among women world- 23.2/100,000, and the ASR of mortality was approximately wide, with an incidence that has displayed a gradual in- 4.9/100,000. Although the incidence of breast cancer creasing trend in many countries over the past 30 years among Chinese women was relatively lower than that in [1]. According to the GLOBOCAN 2012 released by the developed countries, an increasing trend has been wit- International Agency of Research on Cancer (IARC), nessed in recent years [3]. there were approximately 1.7 million newly diagnosed At present, it is widely believed that the development cases of breast cancer and 0.5 million deaths in women of breast cancer can be attributed to genetic factors, life- worldwide in 2012 [2]. Moreover, the age-standardized style changes, and environmental exposure, among rate (ASR) of mortality in developed countries was 1.8 which environmental factors and individual behaviors times that in developing countries [2]. are believed to be factors that can be modified to pre- In China, breast cancer ranked as the most common vent breast cancer [4]. However, the risk factors for type of cancer and the fifth leading cause of cancer deaths breast cancer might be different between the Chinese and Western populations [5]. Therefore, studies are war- * Correspondence: [email protected] ranted to explore the causes of breast cancer in China. † Equal contributors Certain personal characteristics, such as genetic inher- 1Shenzhen Center for Chronic Disease Control, Shenzhen, Guangdong 518020, P. R. China itance and lifestyle, have been explored in a few previous Full list of author information is available at the end of the article studies [6,7], but spatial distribution information is rare. © 2015 Zhou et al.; licensee BioMed Central. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zhou et al. Chinese Journal of Cancer (2015) 34:13 Page 2 of 7 Such analysis will be useful in exploring the risk factors Table 1 Quality evaluation of breast cancer registration associated with the distribution patterns of breast cancer, at each district of Shenzhen between 2007 and 2012 which can provide not only etiologic clues but also District MV% DCO% O&U% decision-making information for the effective implementa- Luohu 93.45% 0.89% 1.85% tion of breast cancer prevention and health promotion. Nanshan 92.57% 0.76% 1.43% This study aimed to explore the spatio-temporal distri- Futian 93.75% 0.72% 1.56% bution pattern of female breast cancer using the cancer information obtained from the Shenzhen Cancer Regis- Yantian 91.57% 1.08% 2.07% try System. Bao’an 89.72% 1.57% 2.93% Longgang 87.48% 1.82% 2.89% Methods Guangming 89.71% 1.37% 2.68% Data source Pingshan 88.45% 1.24% 2.71% The data on the breast cancer cases in this study were Longhua 87.51% 1.37% 2.84% obtained from the Shenzhen Cancer Registry System, which was established in 1998 and covers all permanent Dapeng 86.45% 1.45% 3.04% residents in Shenzhen city. In this system, all of the Total 90.04% 1.25% 2.84% breast cancer cases diagnosed in qualified hospitals (de- MV%, percentage of cases with morphologic verification; DCO%, percentage of cancer cases identified with death certificates only; O&U%, percentage of other fined as the hospitals with tumor diagnosis and treat- and unspecified cases. ment qualifications) were requested to be reported with a unified tumor reporting card according to the Inter- Spatial clusters national Classification of Diseases, 10th revision (ICD- A spatial cluster analysis of breast cancer cases was per- 10). In addition, these data were supplemented by the formed using spatial autocorrelation [9]. A spatial cluster Shenzhen Death Registration System to account for po- model of the overall area was estimated via the global tentially under-reported cases. spatial autocorrelation index Moran’s I (global indicators The incidence of breast cancer was estimated accord- of spatial association, GISA); the cluster type and exact ing to the population data from the statistical yearbook position were examined using local Moran’s I (local indi- of Shenzhen with age groups calculated using the 2010 cators of spatial association, LISA). The values of global census information from Shenzhen City [8]. The data Moran’s I ranged from −1 to 1, and the greater the abso- from the Shenzhen Cancer Registry System between lute correlation value, the stronger the spatial autocor- 2007 and 2012 were included in this study. It should be relation. When I > 0, the disease distribution is positive noted that there was a lag of 1 year for the cancer regis- for spatial autocorrelation and vice versa. A high I value tries to verify and clean the registered data before the (hotspot, high-high) or low I value (coldspot, low-low) data were available for analysis. For each case, informa- exists when the LISA statistics are positive, and different tion on the place of residence was classified according to observations (low-high) are present when the LISA sta- the minor civil division with the ratio of 1:10,000 based tistics are negative. on the geographic information in the Shenzhen adminis- trative division map provided by the National Geographic Spatio-temporal scan Center of China. The spatio-temporal cluster detection test for breast cancer incidence was retrospectively performed using spatial scan statistics. The scan parameters were as fol- Quality control lows: the time range was between 2007 and 2012; the The percentage of cases with morphologic verification time interval was 1 year; the potential population risk (MV%), percentage of cancer cases identified with death was 10%; and the number of Monte Carlo simulations certificates only (DCO%), and percentage of other and was restricted to 999 times [10]. Then, the log likelihood unspecified cases (O&U%) were used to evaluate the ratio (LLR) was obtained from the actual incidence and completeness, validity, and reliability of the cancer regis- theoretical incidence computed by the Poisson distribu- tration data. According to the acceptable criteria of the tion in each scan window. The formula was as follows: C c IARC, the following standards should be reached: an LLR = log(c/n)c [(C - c)/(C - n)]( - ) (where C is the MV% higher than 66%, a DCO% lower than 15%, and an total number of cases, c is the number of cases in the O&U% lower than 5%. scanning window, and n is the expected number of cases The overall values of MV%, DCO%, and O&U% were in the active scanning window). The scanned area in- 90.04%, 1.25%, and 2.84%, respectively. The quality volving the maximum LLR value with statistical signifi- evaluation for each cancer registration is presented in cance was defined as a Class 1 cluster, and the other Table 1. scanned areas containing only LLR values with statistical Zhou et al. Chinese Journal of Cancer (2015) 34:13 Page 3 of 7 significance were identified as Class 2 clusters. The rela- was 20.0/100,000 with a CASR of 29.1/100,000 and a tive risk (RR) was calculated as the ratio of the incidence WASR of 21.1/100,000.